Implement masked view #102
Conversation
martinRenou
force-pushed
the
implement_masked_view
branch
3 times, most recently
from
ce98bfb
to
b441e3b
Compare
martinRenou
force-pushed
the
implement_masked_view
branch
2 times, most recently
from
987380b
to
a2a9418
Compare
martinRenou
force-pushed
the
implement_masked_view
branch
2 times, most recently
from
3b3fe1f
to
dedb521
Compare
martinRenou
force-pushed
the
implement_masked_view
branch
from
dedb521
to
f6c8660
Compare
martinRenou
force-pushed
the
implement_masked_view
branch
from
f6c8660
to
f75518a
Compare
|
what's the equivalent to a masked_view in pandas / xarray? |
|
ok, i see it's the where op in xarray. |
| template <class S> | ||
| disable_integral_t<S, reference> operator[](const S& index); | ||
| template <class I> | ||
| reference operator[](std::initializer_list<I> index); |
There was a problem hiding this comment.
I think we should not make this an initializer_list of any more. I've noted the same for xtensor. There is a cost when it get's casted from int32 to int64 which can be avoided by making it size_t from the start.
There was a problem hiding this comment.
So you want me to remove it completely ? Isn't it a breaking change in xtensor ?
There was a problem hiding this comment.
actually we need to define what we make it i guess.
I think it should take an index_type (i.e. std::array<std::size_t, ... > or svector<size_t>) then we can still easily construct it from an initializer list.
we need to discuss this (anyways, since it is implemented like this in xtensor, we can prob keep it for now :)
There was a problem hiding this comment.
Ok I see :) maybe we should open an issue in xtensor ?
No description provided.